A new framework called HERO has been developed to improve generative recommendation systems by addressing the issue of popularity bias. This framework utilizes heterogeneous teacher distillation, employing two specialized teachers for popular and unpopular items, and routes training samples to the appropriate teacher. Additionally, it incorporates diversity-aware item rollout to ensure that less popular items are also considered during training. HERO has demonstrated effectiveness in offline experiments and, when deployed on the 1688 homepage, resulted in a 3.412% increase in gross merchandise value (GMV) during a 15-day A/B test. AI
IMPACT This framework could lead to more personalized and diverse recommendations, potentially increasing user engagement and sales in e-commerce platforms.
RANK_REASON The cluster describes a new research paper detailing a novel framework for generative recommendation systems. [lever_c_demoted from research: ic=1 ai=1.0]
Read on arXiv cs.IR (Information Retrieval) →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
